test calling duplication

tests/data/make_testdata.sh

@@ -33,3 +33,11 @@ bam="https://ftp-trace.ncbi.nlm.nih.gov/giab/ftp/data/AshkenazimTrio/analysis/10
 region="chr1:115,620,601-115,756,835"
 samtools view "${bam}" "${region}" -o NA24385_10xGenomics_chr1_DEL.bam
 samtools index NA24385_10xGenomics_chr1_DEL.bam
+
+# DUPLICATION ON CHROMOSOME 3 (HCC1954T)
+# GRCh38: chr3  101569210   101663759
+# source: LongRanger calls
+bam="https://s3-us-west-2.amazonaws.com/10x.files/samples/genome/HCC1954T_WGS_210/HCC1954T_WGS_210_phased_possorted_bam.bam"
+region="chr3:101,508,502-101,725,227"
+samtools view "${bam}" "${region}" -o HCC1954T_10xGenomics_chr3_DUP.bam
+samtools index HCC1954T_10xGenomics_chr3_DUP.bam

tests/test_run.py

@@ -11,6 +11,7 @@
 CANDIDATES_INV = Path("tests/data/HCC1954T_10xGenomics_chr21_INV.candidates.bedpe").absolute()
 CANDIDATES_REV_INV = Path("tests/data/HCC1954T_10xGenomics_chr21_INV.candidates_reversed.bedpe").absolute()
 BAM_DEL = Path("tests/data/NA24385_10xGenomics_chr1_DEL.bam").absolute()
+BAM_DUP = Path("tests/data/HCC1954T_10xGenomics_chr3_DUP.bam").absolute()
 BAM_TRA = Path("tests/data/HCC1954T_10xGenomics_chr_12_20_TRA.bam").absolute()
 
 
@@ -195,6 +196,25 @@ def test_call_deletion(tmp_path):
         assert abs(nas_pass[0].break2 - 115690219) < 500
 
 
+def test_call_duplication(tmp_path):
+    config_file = io.StringIO(f"bam_file={BAM_DUP}\noutdir={tmp_path}\nDEBUG=True\n")
+    configs = Configs.from_file(config_file)
+
+    exitcode = run(configs)
+    assert exitcode == 0
+    output = tmp_path / "NAIBR_SVs.bedpe"
+    with open(output) as bedpe_reader:
+        nas = list(iter_novel_adjacencies(bedpe_reader))
+        assert len(nas) > 0
+        nas_pass = [na for na in nas if na.pass_threshold]
+        assert len(nas_pass) == 1
+        assert nas_pass[0].orient == "-+"
+        assert nas_pass[0].chrm1 == nas_pass[0].chrm2 == "chr3"
+        assert abs(nas_pass[0].break1 - 101569210) < 500
+        assert abs(nas_pass[0].break2 - 101663759) < 500
+        assert False
+
+
 def test_call_interchromosomal(tmp_path):
     config_file = io.StringIO(f"bam_file={BAM_TRA}\noutdir={tmp_path}\nDEBUG=True\n")
     configs = Configs.from_file(config_file)